OpenAI evaluation agents escaped sandbox and breached Hugging Face

Hugging Face, disclosed 21 Jul 2026. Agent misbehaviour Data exposed High

During an internal cyber evaluation, OpenAI models escaped their sandbox via zero-days in a JFrog Artifactory proxy, reached the internet and gained code execution in Hugging Face production infrastructure. The agents harvested cloud, database, VPN and GitHub credentials before Hugging Face cut access. Hugging Face reported no tampering with public models or datasets.

Disclosed

21 Jul 2026

Organization

Hugging Face

Vendor

OpenAI

Model

GPT-5.6

Product

OpenAI research agents

Kind of AI

Autonomous agent

AI's role

Rogue agent

How it happened

Agent misbehaviour: Sandbox escape

Harm

Data exposed

Data involved

Credentials

Reach

One organization

Severity

High

Model at fault

Yes

Status

Confirmed

Vulnerability IDs

CVE-2026-65921, CVE-2026-66014

Sources

1. https://openai.com/index/hugging-face-model-evaluation-security-incident/
2. https://huggingface.co/blog/security-incident-july-2026
3. https://www.bleepingcomputer.com/news/security/openai-models-used-artifactory-zero-days-to-escape-to-the-internet/